a dynamic model of top management team effectiveness
TRANSCRIPT
The Leadership Quarterly 14 (2003) 297–325
A dynamic model of top management team effectiveness:
managing unstructured task streams
Amy C. Edmondson*, Michael A. Roberto, Michael D. Watkins
Harvard Business School, Boston, MA 02163, USA
Accepted 28 February 2003
Abstract
Leadership research relating top management team (TMT) demographics to firm performance has
produced mixed empirical results. This article suggests a new explanation for these inconsistencies.
We first note that a given TMT is likely to face a variety of different situations over time. Thus, while
TMT demographic composition is relatively stable, the TMT task is dynamic and variable. In some
situations, team members have similar information and interests (a symmetric distribution); in others,
information or interests diverge (an asymmetric distribution). Based on team effectiveness theory, we
argue that, unless group process is managed accordingly, asymmetric distributions of situation-specific
information and interests will reduce TMT decision-making effectiveness. We then develop leader
process choices to mitigate the potentially harmful effect of these asymmetries. These arguments form
the basis of a theoretical model of TMT effectiveness that integrates insights from research on
leadership, TMTs, small group process, and negotiation, and has practical implications for how leaders
of senior teams can improve team effectiveness through appropriate process choices.
D 2003 Elsevier Science Inc. All rights reserved.
1. Introduction
Encouraging the CEO and senior executives to work as a team has been suggested as a
way of enhancing strategic leadership effectiveness in complex organizations. Through
strategic leadership (Boal & Hooijberg, 2000), an organization maneuvers forward into an
1048-9843/03/$ – see front matter D 2003 Elsevier Science Inc. All rights reserved.
doi:10.1016/S1048-9843(03)00021-3
* Corresponding author. Tel.: +1-617-495-6732; fax: +1-617-496-5265.
E-mail addresses: [email protected] (A.C. Edmondson), [email protected] (M.A. Roberto),
[email protected] (M.D. Watkins).
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325298
imperfectly known future, making commitments to some opportunities while turning away
from others. Many practitioners and scholars have argued that teamwork at the top promotes
the generation of creative ideas and multiple alternatives, enables executives to utilize diverse
experience to solve difficult problems, and increases involvement and commitment of key
senior executives (Ancona & Nadler, 1989; Bauman, Jackson, & Lawrence, 1997; Nadler,
1996). A top management team (TMT) provides a way to cope with the turbulence and
complexity in the external environment that has complicated the task of executive leadership
(Hambrick, 1998; Janofsky, 1993; Nadler, 1998). Teamwork allows the CEO to engage in a
participative group process through which diverse members wrestle together with difficult
issues to make decisions and build commitment to implementing them, giving rise to strategic
leadership effectiveness.
At the same time, considerable research and anecdotal evidence suggest that TMTs often
fail to achieve their potential. Some scholars have found that many senior teams do not
engage in real teamwork (Hackman, 1990; Hambrick, 1994; Katzenbach, 1998). Others have
reported that TMTs can find it difficult to resolve conflict (Amason, 1996), build commitment
(Wooldridge & Floyd, 1990), or reach closure in a timely fashion (Eisenhardt, 1989;
Harrison, 1996; Hickson, Wilson, Cray, Mallory, & Butler, 1986). Several in-depth case
studies document how dysfunctional group dynamics can lead to errors in judgment and
flawed decisions. Notably, Janis’ (1982) early work on groupthink attributed certain foreign
policy fiascoes to the pressures for conformity that arise within cohesive senior groups, and
Ross and Staw (1986, 1993) conducted case studies examining how groups of senior
executives escalate commitment to failing courses of action.
These leadership failures can be explained by an inability to manage group processes
effectively. Team researchers use the term ‘‘process losses’’ (Steiner, 1972) to describe a gap,
attributable to ineffective communication or coordination, between a group’s potential and
actual performance. Potential performance is the level that could be achieved, from a given
set of inputs, such as sufficient resources, diversity of experience, or other advantages. Well-
documented process losses include pressures for conformity that lead to premature conver-
gence on a solution (Janis, 1982; Taras, 1991), group discussion that fails to surface relevant
information members possess (Stasser, 1999), and other coordination and motivation failures.
Although process losses are likely to occur in TMTs—reducing decision quality and overall
team effectiveness—the question of what factors might make senior teams less vulnerable to
them has not been addressed in the literature.
This article develops implications of research on group interaction process for TMTs.
Instead of using relatively stable team characteristics such as demographic composition to
explain differences in outcomes, we focus on the dynamic relationship between the team and
the different situations it faces over time. We propose that the same team composition will have
a different effect on TMT effectiveness depending on situation-specific distributions of
information and interests. Reflecting Hackman (1987), we define TMTeffectiveness as follows:
(1) the degree to which the team’s decisions enhance organizational performance (e.g,
Hambrick, 1994), (2) members’ commitment to implementing team decisions and willingness
to work together in the future (Amason, 1996; Nadler, 1996; Schweiger, Sandberg, & Ragan,
1986), and (3) the extent to which the team process meets members’ growth and satisfaction
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 299
needs (Hackman, 1987; Hambrick, 1994). Situation-specific distributions refer to the distinct-
ive information or interests held by team members in a given situation.
In the next section, we review two streams of literature that inform our understanding of
TMTs and leadership decision-making. Section 3 builds on this background to discuss
implications of the task variability faced by senior teams. Section 4 develops the core of our
theoretical model, which proposes that situation-specific distributions of information and
interests influence TMT effectiveness, moderated by power centralization and psychological
safety. In Section 5, we develop a set of process choices through which leaders can take
actions that alter the effects of situational factors on team effectiveness to produce better
results (see Table 1 in Section 5 for a summary). Section 6 discusses implications of these
arguments for theory and practice.
2. An integrative approach
Two streams of leadership research can contribute to extending theory on TMTs. The first,
strategic leadership research, employs a macro lens to study effects of TMT demographics on
organizational performance, positing that team process mediates this relationship; the second,
supervisory leadership research, employs a micro lens to examine leader decision-making
behavior within a firm (Boal & Hooijberg, 2000). A central focus of the second stream is the
extent to which leaders seek and use input from others in making critical decisions, clearly, an
important issue for understanding TMTs. We review elements of both streams of research as a
foundation for proposing an integrative approach.
2.1. TMT demographics research
The ‘‘upper echelons’’ literature (Hambrick & Mason, 1984) attempts to link the attributes
of the firm’s leaders with strategic choices and organizational outcomes. Using TMT
demographics as proxies for psychological characteristics, researchers relate variables such
as age, tenure, education, and functional background to organizational outcomes such as sales
growth, innovation, and executive turnover (e.g. Bantel & Jackson, 1989; Eisenhardt &
Schoonhoven, 1990; Finkelstein & Hambrick, 1990; Keck & Tushman, 1993; Murmann &
Tushman, 1997). TMT composition also has been used to predict team communication and
conflict (e.g., Knight et al., 1999; Miller, Burke, & Glick, 1998; Papadakis, Lioukas, &
Chambers, 1998; Smith et al., 1994). This stream of research employs an input–process–
output model, focusing predominantly on the relationships among certain inputs and outputs
with less attention to intervening process variables.
Although studies have found relationships between demographic variables and outcomes,
the data have produced conflicting results. For example, Murray (1989) found that tenure
heterogeneity was associated with higher performance among firms in the oil industry;
Eisenhardt and Schoonhoven (1990) found that tenure heterogeneity was positively asso-
ciated with revenue growth for semiconductor firms, and Murmann and Tushman (1997)
discovered that tenure heterogeneity was associated with faster responses to environmental
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325300
change in the cement industry. At the same time, other research found that tenure
heterogeneity is negatively associated with firm performance (Williams & O’Reilly, 1998).
Hambrick, Cho, and Chen (1996) showed that heterogeneous teams responded more slowly
to competitors’ actions in the airline industry, Wagner, Pfeffer, and O’Reilly (1984) reported
that teams with high tenure heterogeneity experienced more turnover, and Smith et al. (1994)
found that tenure heterogeneity was associated with lower returns on investment in a sample
of 53 high-technology firms.
Faced with these inconsistencies, scholars have identified three important limitations of the
TMT literature. First, demographic characteristics are, at best, imperfect proxies for psycho-
logical constructs (Boal &Hooijberg, 2000). Second, TMT research pays ‘‘too little attention to
the actual mechanisms that serve to convert group characteristics into organization outcomes’’
(Hambrick, 1994, p. 185). Research thus has not explained how and why certain demographic
attributes affect outcomes, nor provided definitive conclusions regarding the direction of
causality (Bower, 1998; Hambrick, 1994; Pettigrew, 1992). Most studies assume group process
as a mechanism without measuring it directly. In a rare exception, process was measured and
was a stronger predictor of performance than demographic composition (Smith et al., 1994);
nonetheless, enthusiasm for composition as a proxy for process remains strong because of the
relative ease of obtaining demographic over process data.
Third, most demographic studies fail to account for the impact that situation-specific factors
have on team process and performance (Papadakis et al., 1998). Clearly, TMTeffectiveness can
vary from one situation to another (Janis, 1982; Katzenbach, 1998), yet TMT composition
changes infrequently. Demographic analysis therefore necessarily provides an incomplete
explanation of variation in a team’s performance over time. This suggests a need for additional
theory to explain how situational factors and team attributes work together to shape TMT
processes and outcomes.
2.2. Situational leadership research
Situation-to-situation differences are the focus of a substantial body of research on
leadership behavior (Evans, 1970; Fiedler, 1967; Hersey & Blanchard, 1969; Vroom&Yetton,
1973). This work suggests that leaders need to adjust their style or approach based upon the
circumstances and conditions that they encounter in their organizations and environments.
Normative decision theory (Tannebaum & Schmidt, 1958; Vroom & Jago, 1988; Vroom &
Yetton, 1973) focuses specifically on how leaders make decisions in conjunction with their
management teams, and suggests that leaders ought to interact differently with these
subordinates based upon situational attributes. Vroom and Yetton (1973) argue, in particular,
that leaders should invite more or less subordinate participation during a decision-making
process depending upon situational characteristics such as the importance of the quality of the
decision, the level of time pressure, and the extent to which subordinate commitment is critical
to successful implementation. Leaders’ process choices in this model range from highly
directive (making the decision without input from subordinates) to highly participative
(working with subordinates to develop and evaluate alternatives and then reach consensus
on the final decision).
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 301
Empirical studies have found substantial support for normative decision theory’s proposi-
tions (Field, 1982; Vroom & Jago, 1988), suggesting that a situation-contingent approach to
TMT decision-making processes is worth pursuing. Because normative decision theory deals
only with the issue of whether to utilize a team to make a decision—not how to manage a team
process to produce optimal outcomes—its propositions do not help leaders avoid the process
losses that groups encounter on a regular basis (Steiner, 1972). Moreover, normative decision
theory focuses on characteristics of the situation or decision itself, rather than examining how
certain situation-specific attributes of the team might affect a leader’s process choices.
2.3. Summary
Upper echelons research focuses on TMT composition as the critical predictor of
effectiveness. Its empirical studies rely on relatively stable and deterministic causal models
that ignore the ways team processes and outcomes might vary across the multiple situations
faced by senior teams. This work thus assumes a consistency of conditions and team
performance that is unlikely to exist in real TMTs, and it underspecifies the role of process.
Normative decision theory, in contrast, prescribes leader choices contingent on the
situation. However, this work has focused on whether or not a leader should employ a team,
rather than how to employ different kinds of team processes in different situations.
This paper proposes an integrative approach, in which TMT effectiveness depends both on
team composition and on how the team leader manages team process to reflect situational
factors. In the next section, we thus identify critical characteristics of TMTs that vary with the
situation and address implications of this variability for group processes and outcomes.
3. Situation-specific asymmetries
Psychological research on small groups can inform our understanding of TMT process.
Like many decision-making groups (e.g., Hollenbeck, Ilgen, LePine, Colquitt, & Hedlund,
1998; Hollenbeck et al., 1995), TMTs are hierarchical and have distributed expertise. For
example, product development teams often have strong leaders and highly diverse expertise
(Clark & Wheelwright, 1992; Lewis, Welsh, Dehler, & Green, 2002). At the same time,
scholars recognize the unique nature of TMTs and pay special attention to attributes that
distinguish them from work groups lower in the organizational hierarchy (Hambrick, 1994;
Nadler, 1998). We thus highlight salient characteristics of TMTs to build a foundation for
developing theoretical propositions to explain process failures in these groups.
3.1. Unstructured task streams
TMTs face ambiguous and ill-structured problems (Ancona & Nadler, 1989; Hambrick,
1994). Many scholars have observed that senior teams perform tasks that are more complex
and unstructured than the activities carried out by most other organizational work teams
(Ancona & Nadler, 1989; Hambrick, 1994; Nadler, 1998). Senior teams must comprehend
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325302
and interpret a great deal of vague, ambiguous, and often conflicting, information from many
sources (Hambrick, 1994). They manage diverse external constituents, from the Board of
Directors, to shareholders, analysts, government officials, and potential alliance partners
(Ancona & Nadler, 1989; Hambrick, 1994; Nadler, 1998). Furthermore, TMTs must decide
which (of many possible options) are the most critical tasks to perform, unlike most work
groups, which are assigned to carry out specific tasks.
We identify an additional dimension of task complexity. Senior teams confront a wider
variety of ‘‘tasks’’ (that is, situations, issues, and challenges) than other teams. The TMT task
is inherently variable. In contrast, most organizational teams, including both self-managed
and hierarchical distributed expertise (HDE) teams, tend to undertake relatively structured,
unitary tasks that exhibit some degree of consistency over time (Hackman, 1987, 1990;
Hollenbeck et al., 1995, 1998). We conceptualize this task variability as follows: senior teams
face unstructured task streams—a continual flow of varying and overlapping situations. In
these streams, some situations may be familiar and routine, while others demand substantial
investments in problem definition or creation of new knowledge. For consistency, we use the
term ‘‘situation’’ to refer to the task, issue, or decision confronting a TMT at a given time.
3.2. Dynamic distributions of information and interests
The dynamic nature of this unstructured task stream implies that the team–situation
relationship—or the match between the team’s stable characteristics and the situation at
hand—will vary across time. We identify two core dimensions of the team–situation
relationship.
First, the distribution of relevant information within the team is likely to differ across
situations, depending on the relationship of a given issue to members’ current activities and
functional expertise. Situation-specific information consists of facts, data, and ideas that are
pertinent to a particular decision.
Second, TMT members, as representatives of powerful constituencies within the organ-
ization, may have closely aligned interests on some situations and divergent or competing
interests on others. Situation-specific interests comprise goals and objectives that individual
team members wish to achieve, sometimes at the expense of other team members. The nature
and distribution of interests within the TMT thus differs from one situation to another.
Distributions of information and interests are a central focus of negotiation research (Lax
& Sebenius, 1986; Raiffa, 1982; Walton & McKersie, 1965; Watkins, 2000). This work is
particularly relevant for TMTs because decision-making in these groups often resembles
multiparty, mixed-motive negotiations rather than collaborative problem-solving processes
(Allison, 1971; Bazerman, 1998; Cyert & March, 1963; Murray, 1978). Negotiation scholars
contend that assessing how information and interests are distributed among parties is critical
to understanding multiparty negotiation processes (e.g., Raiffa, 1982). In some situations,
participants enter a negotiation with common information about each person’s best alternative
to a negotiated agreement or ‘‘BATNA’’ (Fisher & Ury, 1991); hence, each knows others’
walk-away points. Scholars describe this distribution of information as symmetric. In other
situations, each negotiator knows his or her own BATNA but is uncertain of others’—an
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 303
asymmetric distribution. Likewise, individuals in a negotiation may have substantially
aligned (symmetric) interests, or they may have substantially opposing (asymmetric) interests
(e.g., Lax & Sebenius, 1986; Watkins, 2000).
The state of a TMT relative to a specific situation can also be characterized along these
dimensions. Each member of a TMTmay have the same situation-specific information, or they
may have access to private or unique information not possessed by others. We define TMT
information asymmetry as the degree to which different team members have distinct, unshared
information about a particular situation. For example, facing a decision about a merger, team
members are likely to have different information about the strategy, organization, and finances
of potential partners. In many cases, the relevant information may be so taken for granted by
individual team members that they are unaware of others’ lack of knowledge or understanding
(Argyris, 1993; Larson, Foster-Fishman, & Keys, 1994).
Similarly, interests of teammembers may be strongly aligned, or highly divergent.We define
TMT interest asymmetry as the degree to which team members have divergent interests in a
given situation. For example, facing a decision to downsize the workforce, team members may
wish to preserve their own power and resources and therefore try to minimize layoffs in their
respective areas. In such ‘‘mixed-motive’’ situations, team members may both cooperate to
accomplish joint objectives and compete to advance individual interests (Bazerman,Mannix, &
Thompson, 1988; Lax & Sebenius, 1986; Walton & McKersie, 1965).
These asymmetries are more dynamic and situation-dependent than demographic variables
in teams (such as tenure heterogeneity), which tend to be relatively stable over time. The same
team thus may have a symmetrical distribution of information and/or interests for one
situation and an asymmetrical distribution for another. For instance, member interests may be
closely aligned in a decision about fending off an unsolicited takeover bid, while interests may
be quite divergent when executives gather to determine the annual budget at a time when
resources are limited. This suggests that models that rely on stable compositional variables
may miss an important source of variance in team effectiveness, as depicted in Fig. 1.
Fig. 1. Situation variability creates situation-specific asymmetries affecting TMT process and effectiveness.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325304
4. Diagnosing process losses
We draw from research on small groups to suggest that asymmetries in the distribution
of situation-specific information and interests may lead to process losses in TMTs. One
research tradition views groups through a computational lens, focusing on how inputs
(information that individual members possess) are transformed into outputs (group
decisions). Decision outcomes are a function of both the a priori distribution of information
and the efficiency with which a group surfaces and uses its distributed information
(Hollenbeck et al., 1998; Stasser, 1999; Stasser & Davis, 1981). Another tradition stems
from the psychology of social influence processes (e.g., Asch, 1951; Baron, Vandello, &
Brunsman, 1996; Hovland & Weiss, 1951; Janis & Mann, 1977; Zimbardo & Leippe,
1991) and employs a socio-emotional lens. In this work, outcomes are a function of
dominance patterns in group discussions (Bales, 1954), suppression of dissent (Janis, 1982),
or whether and how minority viewpoints are considered (Nemeth, 1997; Nemeth &
Wachtler, 1974). Both traditions document process failures, with the former focused on
poor use of information and the latter on failure to consider the full range of ideas held by
group members.
We use both conceptual lenses to analyze implications of distributions of informa-
tion and interests in TMTs for group process and performance. Two relatively stable
TMT attributes—power centralization and psychological safety—are explored as moder-
ators of the relationships between dynamic situation-specific asymmetries and TMT
effectiveness.
4.1. Information asymmetry
One purpose for engaging in a team decision process is to pool expertise from multiple
sources and thereby generate ideas that no individual could develop alone. In this way,
leaders may anticipate capturing synergistic benefits from teamwork. Such benefits often
remain elusive, however. First, teamwork is not automatic (Hackman, 1998). Although
asymmetric information promises synergy from integrating multiple perspectives and
diverse expertise, it also can interfere with team functioning (Bantel & Jackson, 1989).
Compared to having the same information and experience, team members with divergent
information may bring taken-for-granted assumptions and competing mental models that
inhibit mutual understanding (Cannon-Bowers, Salas, & Converse, 1993; Dougherty,
1992). Seeing a situation through a lens partly shaped by the information each holds,
group members may misinterpret each other’s comments.
Second, unique information (known by only one member) tends not to be shared in
group discussions (Stasser, 1999). Experimental studies have demonstrated that groups
tend to dwell on common information (that held by all members), such that privately
held information fails to surface; further, when it does surface, its impact is often muted
(Larson, Christensen, Abbott, & Franz, 1996; Stasser, 1999; Stasser & Titus, 1985). Note
that this phenomenon is not dependent on interests being asymmetric. Private information
may remain unshared when individuals—deeply engaged in the discussion at hand—fail
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 305
to recognize its salience for the issue under consideration. Members also may fail to
share private information because they take it for granted and implicitly assume that
others know what they know, or because they are reluctant to jump into an already active
discussion.
If situation-relevant information fails to surface in a TMT decision-making process, the
quality of outcomes will be reduced. By not discussing all pertinent information, teams
may overlook plausible options, fail to examine the full consequences of each alternative,
or underestimate the risks associated with a proposal. In addition, members’ awareness
that relevant information did not surface is likely to erode commitment to implementing
the group’s decision, especially if they feel they did not contribute fully to it. In contrast,
if members have relatively more situation-relevant information in common at the outset,
the group will more effortlessly consider that information (Stasser, 1999).
Proposition 1: Situation-specific information asymmetry will reduce TMT effectiveness.
This problem is likely to be particularly acute in TMTs characterized by power
centralization. We define power as the capability of an actor to influence others’ behavior
and to get people to do what they otherwise would choose not to do (Finkelstein, 1992;
Kanter, 1979; Pfeffer, 1981, 1992). Scholars have argued that power plays a more central
role in strategic decision-making within top teams than in the tasks performed by other
work groups (Child, 1972; Finkelstein, 1992; Finkelstein & Hambrick, 1994; Mintzberg,
1983; Tushman, 1977). TMTs must make strategic choices even when cause-and-effect
relationships are unclear and when members’ goals are ambiguous and conflicting, while
lower-level groups tend to work on problems characterized by goal agreement and a more
complete understanding of cause–effect relationships (Child, 1972; Tushman, 1977) in
contexts conducive to sharing power and leadership (Avolio, Jung, Murry, & Sivasubra-
maniam, 1996; Seers, 1996).
Power centralization in a TMT is likely to influence the sharing of private information.
Researchers have shown that some TMTs consist of individuals with roughly equal power,
such that the CEO does not maintain a substantial advantage over other members; in others,
the CEO has a great deal more power than the rest of the team (Eisenhardt & Bourgeois,
1988; Finkelstein, 1988, 1992).
We hypothesize that power centralization—that is, when the CEO has a great deal more
power than other members—will inhibit open sharing of private information. Group
members with less power often defer to those with more power (Bales, 1988; Maier,
1961; Pfeffer, 1992; Russo & Schoemaker, 2002), in part because powerful group
members can reject or marginalize those holding minority views (Nemeth, 1997; Nemeth
& Wachtler, 1974; Schachter, 1951). Less powerful members may engage in self-
censorship with respect to dissenting views (Janis, 1982) while more powerful members
withhold private information to protect and increase their power (Pettigrew, 1973).
Eisenhardt and Bourgeois (1988) found that TMTs with a high degree of power
centralization engaged in less candid discussion and less open exchange of ideas than
teams with more balanced power distributions. When power was centralized, team
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325306
members worked behind-the-scenes to influence the final outcome, such as withholding
critical data or distributing it selectively to others (Eisenhardt & Bourgeois, 1988). We
predict that power centralization in a TMT will moderate the relationship between
information asymmetry and effectiveness as follows:
Proposition 2: Power centralization increases the negative effect of situation-specific
information asymmetry on TMT effectiveness.
Next we suggest that the level of psychological safety in a senior team will affect the
relationship between information asymmetry and team effectiveness. Team psychological
safety is defined as the shared belief that the team is safe for interpersonal risk-taking
(Edmondson, 1999). In teams with high psychological safety, a function of interpersonal
trust and mutual respect, members believe that the group will not rebuke, marginalize, or
penalize them for speaking up or for challenging prevailing opinion.
Psychological safety is salient (and often low) in TMTs for several reasons. First, the
CEO has the ability to hire, fire, and set compensation for the other team members. This
can make it difficult for other senior executives to take interpersonal risks, because they
fear the consequences of missteps as evocatively depicted in one qualitative study
(Edmondson, 2002). Second, conflicts and disputes within senior teams often become
public knowledge. This external visibility creates an added pressure that may affect the
climate within a TMT. Moreover, TMT members can find themselves competing with
one another in a highly politicized and public contest for CEO succession. The stakes are
high, not simply because the winner becomes the chief executive, but also because the
losers typically must search for new employment. The nature of these high-stakes
contests can make it difficult to develop a safe environment for open discussion.
Power centralization and psychological safety are not uncorrelated. Teams with greater
power centralization are less likely to perceive the environment as safe for interpersonal
risk. However, teams can exhibit high psychological safety despite substantial differences
in power among members (Edmondson, 2002; Edmondson, Bohmer, & Pisano, 2000,
2001), or low psychological safety with relatively balanced power structures (Edmondson,
1999, 2002). This is because powerful leaders can take actions that create psychological
safety, such as acknowledging their own mistakes, actively inviting others’ ideas, and
communicating a genuine interest in open discussion and experimentation.
In teams characterized by a low level of psychological safety, individuals may feel
uncomfortable revealing uniquely held information. They may become preoccupied with
concerns about the risk of sharing information and particularly reticent to provide
information that does not confirm existing views within the group. In contrast, individuals
are more likely to share private information if psychological safety is high. Thus, we
propose that psychological safety will moderate the relationship between information
asymmetry and team effectiveness as follows:
Proposition 3: Psychological safety decreases the negative effect of situation-specific
information asymmetry on TMT effectiveness.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 307
4.2. Interest asymmetry
TMTs may experience process losses when discussing issues for which members have
divergent interests. Although senior executives strive to achieve common goals for the firm,
they also represent powerful subunits or constituencies within the organization. This creates
‘‘a tension regarding group identities’’ and enhances the likelihood of goal conflict and self-
interested behavior in some, but not all, situations (Hambrick, 1994, p. 176). For instance,
members can find themselves in direct competition regarding the allocation of resources
(Bower, 1970). Members of lower-level work groups do not represent such large
constituencies and do not have as many organizational resources under their command
and are therefore less likely to experience identity tension (Hambrick, 1994). As noted
above, such tensions may be exacerbated by succession scenarios, in which members
jockey for advantage in the battle to become CEO.
The negotiation literature distinguishes between ‘‘value-creating’’ and ‘‘value-claiming’’
behaviors that arise when interests are not aligned completely (Lax & Sebenius, 1986).
Value-creating behavior consists of finding ways to advance compatible interests or to
devise mutually beneficial trades that benefit the organization as a whole. However, when
interests are not completely aligned, individuals are motivated to capture or ‘‘claim’’ as
much value as they can (Lax & Sebenius, 1986). In these situations, competitive value-
claiming behavior often takes place at the expense of cooperative value-creating behavior.
We argue that when interests within a top team differ, value-claiming behavior is likely
to decrease the generation of creative new options, as executives become ardent, even
overcommitted, advocates for their positions. Advocates’ self-serving behavior can under-
mine efforts to advance shared goals, preclude the kind of thorough analysis needed to
arrive at the best solution for the company as a whole, and lead to erosion of team
relationships. For a vivid example of this phenomenon, consider the demise of investment
banking firm, Lehman Brothers, in the aftermath of value-claiming behavior by senior
executives with divergent interests (Bazerman, 1998; Lax & Sebenius, 1986; Mannix,
1989). In the 1980s, the firm’s two business units became embroiled in a dispute over the
distribution of profits; each unit advocated a strategy that satisfied its own interests, leading
to the departure of the firm’s respected chairman and to financial distress (Bazerman, 1998;
Lax & Sebenius, 1986). Note that this phenomenon is not dependent on an asymmetric
distribution of information. Differences in interests can lead to self-serving behavior, even
when people know what others know (Bazerman & Neale, 1992).
Excessive value-claiming behavior is likely to undermine TMT effectiveness by inhibiting
the potential for mutual gains, thereby leading to suboptimal solutions for the organization.
When value claiming occurs at the expense of value creation, teams are also likely to lack
commitment to implementing resulting decisions and may be less motivated to work
together in the future, a key aspect of TMT effectiveness. Because individuals perceive
themselves as ‘‘winners’’ or ‘‘losers’’ in these situations, the losers lack commitment to the
final solution, necessarily eroding shared commitment. Finally, interpersonal conflict often
emerges as a result of value-claiming behavior, eroding commitment and group harmony
(Amason, 1996; Jehn, 1995).
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325308
Proposition 4: Situation-specific interest asymmetry reduces TMT effectiveness.
Power centralization exacerbates value-claiming behavior in mixed-motive situations.
Negotiation research has shown that dyads with different interests and unequal power
balances tend to arrive at inferior outcomes relative to those with equal power balances,
because those with more power stand to gain more than those with less, and they impede the
creative processes that might have yielded greater total gains by imposing their preferences
on others (e.g. Mannix & Neale, 1993; McAlister, Bazerman, & Fader, 1986; McClintock,
Messick, Kuhlman, & Campos, 1973). Conversely, scholars have demonstrated that multi-
party negotiations among people with equal power tend to result in more value creation than
negotiations among parties with different levels of power (Mannix, 1993), for several reasons.
First, a powerful CEO may believe that he or she can impose a preferred outcome quickly
and forcefully. Doing so may suppress creative problem solving that could lead to a better
solution. Second, less powerful members may focus primarily on protecting their own
interests in these situations, because they fear that others will try to claim extra value through
the behind-the-scenes lobbying that often occurs within TMTs with highly centralized power
(Eisenhardt & Bourgeois, 1988). Third, given these concerns, less powerful members may be
less willing to cooperate with others to discover mutually beneficial alternatives, approaching
the discussion competitively instead of cooperatively (Mannix & Neale, 1993). Thus, we
propose that power centralization moderates the relationship between interest asymmetry and
TMT effectiveness.
Proposition 5: Power centralization increases the negative effect of situation-specific interest
asymmetry on TMT effectiveness.
Greater psychological safety, in contrast, may enhance value creation in mixed-motive
situations. To identify opportunities for mutual gains or creative new alternatives (Fisher &
Ury, 1991), team members must be willing to come forward with novel proposals and to be
open about their own interests and objectives, as well as to engage in a candid discussion of
each proposal’s costs and benefits for all of the parties involved in the decision process. Frank
dialogue requires a climate in which people have few concerns about being embarrassed or
punished for offering unorthodox suggestions; otherwise, they are likely to focus on
defending their own interests rather than on identifying opportunities for mutual gains. Thus,
psychological safety should moderate the relationship between interest asymmetry and team
effectiveness.
Proposition 6: Psychological safety decreases the negative effect of situation-specific interest
asymmetry on TMT effectiveness.
4.3. The interaction between interest asymmetry and information asymmetry
When interests and information asymmetries occur simultaneously, the likelihood of
team effectiveness decreases further. In these instances, individuals may withhold unique
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 309
information because they wish to utilize it to create a personal advantage during bargaining
with other group members, or because full disclosure may harm their negotiating position.
An individual may hold back information that would enable others to determine his or her
goals or preferences, referred to as the ‘‘strategic’’ use of private information (Bazerman,
1998; Lax & Sebenius, 1986). Thus, we argue that these two variables interact to
exacerbate process losses created by each asymmetry individually. Interest asymmetries
mean that team members are motivated to withhold certain information strategically, while
information asymmetries make it more difficult for people with different interests to
recognize creative opportunities to realize mutual gains, increasing the likelihood of flawed
and incomplete debate.
Proposition 7: Situation-specific interest and information asymmetry interact to decrease
TMT effectiveness.
4.4. Summary
The above analysis identifies specific process losses likely to reduce TMT effectiveness.
Situation-specific asymmetries in information and interests are proposed as causes of process
losses, and hence, ineffective decision-making. This is not to suggest that some well-
functioning groups cannot capitalize on the potential benefits of asymmetry, thereby creating
novel and creative solutions to problems; we simply argue that teams are more likely to
encounter process losses when these asymmetries exist. Fig. 1 models the core argument
developed thus far, suggesting that unstructured task streams give rise to situation-specific
asymmetries of information and interests that affect TMT process and, in turn, TMT
effectiveness.
Because power centralization and concerns about psychological safety can be particularly
acute in TMTs, we addressed the potential moderating effects of these variables. In this way,
we relate reasonably stable attributes of top teams to dynamic (situation-specific) factors to
explain TMT effectiveness.
The next section explores leadership process choices that may mitigate effects of
information and interest asymmetries. First, we note that the team leader, as CEO, has
the positional authority to manage team process. Just as scholars of negotiation argue that
bargaining strategies should be contingent upon the distribution of information and interests
(Fisher & Ury, 1991; Lax & Sebenius, 1986; Raiffa, 1982; Watkins, 2000), we suggest that
leaders can design and lead TMT decision-making processes contingent upon situational
assessments of the distributions of information and interests. We thus develop a prescriptive
contingency model (e.g., Elangovan, 1995; Vroom & Yetton, 1973) for leading TMTs.
5. A prescriptive model of contingent process choices
Drawing from both normative decision theory and negotiation theory, in this section we
propose a prescriptive model for TMT decision-making to minimize the process losses
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325310
described above. The purpose of the model is to articulate leadership strategies for realizing
the promise of teamwork despite input asymmetries. By managing process in a contingent
manner, we argue that senior teams can increase their potential for integrating a mix of
perspectives to produce better outcomes than individual leaders could produce alone. Table 1
summarizes our process choice arguments.
Table 1
Matching the potential for situation-specific process losses with leader’s process choice
Team–situation
variable
Process failure Leader’s process
choice
Behavioral attributes
of high level of
process choice
Outcome of process
choice for team
effectiveness
Interest
asymmetry
Value-claiming
behavior reduces
the potential
for group value
creation or joint
gains; affective
conflict
Outcome control
(high vs. low)
Leader decides
final outcome:
. imposes decision
on group after
deliberations are
complete
Decision outcome
likely to create
most value for
organization as a
whole; affective
conflict reduced
Information
asymmetry
Relevant information
fails to surface in
group discussion—
not motivated by
personal gain but
rather by failure to
recognize salience
or reluctance to jump
into discussion process
Process
intervention
(high vs. low)
Leader intervenes
actively and
frequently in the
discussion to
. facilitate sharing
of situation-
specific
information
. clarify others’
contributions;
emphasizes
private
information
. inquire into
views of silent
group members
All situation-specific
information is
revealed and
discussed by the
group in the
deliberation process;
team commitment
to solution is
enhanced
Interaction
effect
Self-serving behavior
can exacerbate
information-surfacing
failures, as people
deliberately withhold
information to enhance
their own power,
further reducing amount
of relevant information
shared and inhibiting
potential for novelty
and synergy that
produce joint gains
Process design
(high vs. low)
Leader imposes a
structured process
to ensure debate
and thorough
consideration of
alternatives, e.g.,
. uses subgroups
to develop
and debate
alternatives
. assigns one or
more devils’
advocates
A healthy debate
between more
than one alternative
takes place, including
a comprehensive
analysis of issues,
improving decision
quality and
increasing team
commitment
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 311
We do not simply provide prescriptions for reducing or eliminating information and
interest asymmetries for two reasons. First, part of the rationale for teams is to bring
different knowledge, expertise, and interests together; amidst this heterogeneity lies the
promise of team synergy. Second, these asymmetries are dynamic (or situation-specific),
such that a solution to align interests in one situation may not solve the problem in
another.
The model includes three process choices, each addressing one of the three basic
failures described above. By articulating three distinct process choices for leaders, we
depart from prior decision-making research that treats level of subordinates’ contribution
as a single process choice (e.g., Hollenbeck et al., 1998; Tannebaum & Schmidt, 1958;
Vroom & Yetton, 1973). Much of the leadership and team literature implicitly frames
the decision-making task as the job of the leader, with or without others’ input and
with varying degrees of weight assigned to others’ input. For example, researchers
studying decision-making in HDE teams focus on how much weight a team leader places
on other members’ input when making a decision (Hollenbeck et al., 1998). Similarly,
normative decision theory (Vroom & Jago, 1988; Vroom & Yetton, 1973) postulates that
leaders traverse a single spectrum from less to more participative modes of decision-
making.
We extend this work by drawing on Nadler’s (1996) view of executive team leadership
and Elangovan’s (1995) model of managerial dispute resolution. These models both
suggest that leaders can be more or less directive about more than one aspect of a
situation, independently. Nadler argued for a distinction between being directive about
content (what decision is made) and being directive about process (how the decision is
made). Elangovan similarly distinguished between outcome control and process control—
both exercised by a third party engaged in dispute resolution.
We articulate three process choices that a leader of a TMT faces: (1) how to reach
closure on a decision (outcome control), (2) how to facilitate group discussion (a relatively
unobtrusive form of process control), and (3) how to structure debate (a heavy-handed form
of process control, which we call process design). Outcome control provides a way to
reduce process losses caused by interest asymmetry; process control mitigates information
asymmetry, and process design reduces problems caused by the interaction effect of both
asymmetries. This way of modeling the leader’s decision space resembles but does not go
as far as models of shared leadership, in which the leadership function is dynamically
transferred from the leader to other members (Avolio et al., 1996). Compared to leaders of
most work teams, CEOs have far greater positional power and legal and fiduciary
accountability, making truly shared leadership difficult to achieve in these teams; however,
this does not preclude CEOs from influencing team processes to generate better discussion
and better results.
5.1. Outcome control
When trying to reach closure in a decision-making process, the TMT leader faces a choice
about whether to exercise more or less control over the outcome. Exercising outcome control
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325312
as a leader involves asking one’s team to generate and discuss alternatives and then making
the final decision alone. Low outcome control means inviting the team to reach a consensus
decision. High outcome control suggests the leader acts as an arbitrator—listening to
competing arguments and selecting the course of action that he or she believes is best for
the organization. With low outcome control, the leader acts as a mediator—trying to bring
team members with different views together to arrive at a mutually acceptable solution (Lax
& Sebenius, 1986).
We propose that low outcome control works best when a team has symmetric interests.
Low outcome control encourages creative problem solving, in part by communicating an
implicit message that everyone’s views matter. Conversely, if a team has symmetrical
interests but knows the leader will make the decision alone anyway, motivation to participate
in the team discussion may be diminished, potentially eroding the quality of the outcome.
Further, if people have not contributed to a decision, they may lack commitment to its
implementation (Kim & Mauborgne, 1997; Korsgaard, Schweiger, & Sapienza, 1995;
Shapiro, 1993; Spreitzer, 1995). Finally, participating in a decision leads members to view
the process as fair, increasing commitment (Shapiro, 1993). Thus, when interests are
symmetrical and problem solving is unlikely to be impeded by value claiming, leaders can
reach closure through consensus building without excess risk of group process losses.
In contrast, when substantial interest asymmetry exists for a given situation, the leader can
mitigate the harmful effects of value-claiming behavior through increased outcome control.
Knowing that team members may push for decisions that meet their own interests at the
expense of organization-wide interests, the leader can communicate that he or she will make a
final decision. Through outcome control, the leader diminishes the likelihood that a struggle
to reach consensus will result in a compromise that is not best for the firm. The leader also
prevents group members from engaging in a competitive battle that results in one subset of
the TMT imposing its will on others. Through outcome control, the leader also can preclude
problem avoidance (Elangovan, 1995, 1998).
Self-serving behaviors in a team have been shown to promote affective conflict (Amason,
1996; Jehn, 1995; Roberto, 2000) and reduce commitment to implementation (Amason,
1996; Garvin & Roberto, 2001). Outcome control allows the leader to ‘‘call the question,’’
bringing deliberations to a close before affective conflict builds up. Team members are
likely to view the leader making the decision as more fair and appropriate than a process
characterized by self-serving or political behavior on the part of group members in a
difficult consensus building session. Dispute resolution research shows that an orderly
process among parties with conflicting interests enhances perceptions of fairness and
commitment (Folger & Konovsky, 1989; Karambayya & Brett, 1989). Thus, by formulating
a decision outcome after hearing others’ views, the leader can promote the decision quality
and implementation commitment integral to team effectiveness. In sum, leader process
control is likely to moderate the negative effect of situation-specific interest asymmetry on
team effectiveness.
Proposition 8: Outcome control by the leader reduces the negative effect of situation-specific
interest asymmetry on TMT effectiveness.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 313
5.2. Process control
We define process control as leader intervention in the discussion to encourage certain
people to share information before or more often than others, to inquire into the views of
silent members, or to emphasize particular remarks made by members. This may include
managing participation (‘‘air time’’), asking people where they stand on particular issues, or
encouraging alternative views. High process control entails reiterating or paraphrasing points
that surfaced quickly but failed to receive sufficient attention (Larson et al., 1996; Larson,
Foster-Fishman, & Franz, 1998), as well as questioning and testing for understanding
(Nadler, 1998). These leadership behaviors constitute a more directive approach to facilitating
group process (Nadler, 1998; Schwartz, 1994; Webne-Behrman, 1998) relative to low process
control—which involves encouraging a discussion in which members participate as they
wish, refraining from calling on particular individuals and not emphasizing or paraphrasing
others’ comments.
When leaders choose low process control, they allow people to speak freely, rather
than choosing to steer the discussion and select who participates at what time. This
distinction draws from two streams of work: research on groups as information systems
(Larson et al., 1998) and models of process consultation (Schwartz, 1994; Webne-
Behrman, 1998) that provide helpful detail about how to facilitate group discussion for
effectiveness.
We propose that team distribution of information should determine the leaders’ degree of
process control. For a symmetric distribution, high process control by the team leader is
unnecessary and may even be counterproductive. First, it may create discomfort for those
who do not like to speak without having thought through their comments carefully. Second, it
may create a perception that the leader is trying to slant the discussion in a particular
direction, leading a group to abandon potentially superior options they believe the leader
disfavors. Indeed, research has shown that group members may show excessive deference if a
leader reveals his views in the early stages of a team discussion (Levine, 1989). If the
rationale for selecting people to speak is unclear, leader intervention may decrease
perceptions of procedural fairness. Finally, if everyone in the team has the same information
at the outset, encouraging people to reveal what they know is superfluous, such that a less
directive approach is intrinsically preferable.
When the distribution of situation-specific information in a TMT is asymmetric, the risk of
private information failing to surface is high (Stasser, 1999) but can be reduced if the leader
takes an active role in encouraging people to share information. The status of leaders gives them
the ability to encourage others to reveal private information (Larson et al., 1996; Stasser, 1999).
Consistent with this proposition, laboratory studies show that groups with unshared (asym-
metric) information make better quality decisions when leaders intervene actively in the
discussion to emphasize and repeat previously unshared information (Larson et al., 1998).
Encouraging people to share private information also builds commitment, because
individuals feel they have had ample opportunity to influence the final outcome (Shapiro,
1993; Thibault & Walker, 1975). This contention is supported by experimental research
showing that when leaders ask clarifying questions, probe for further explanation, and
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325314
rephrase comments to insure that they have understood people correctly, they foster higher
levels of commitment (Korsgaard et al., 1995); we add that this is particularly relevant when
information asymmetries exist. In this way, process control can moderate the negative effect
of information asymmetry on team effectiveness.
Proposition 9: Process control reduces the negative effect of situation-specific information
asymmetry on TMT effectiveness.
5.3. Process design
Finally, the leader faces a choice about how to design a decision-making process to
ensure healthy debate. The exercise of process design may involve dividing the team into
two subgroups to develop alternatives before coming together to debate the merits of each,
a process choice called the ‘‘Dialectical Inquiry’’ method (Priem, Harrison, & Muir, 1995;
Schweiger et al., 1986). Or, a leader can assign an individual to observe and critique
alternatives being discussed—the ‘‘Devil’s Advocacy’’ Method (e.g., Priem et al., 1995;
Schweiger et al., 1986). These process designs have been proposed as a way to insure that
team members have an opportunity to develop and express their views free from the
pressures to conform to a dominant position within a larger group (Janis, 1982; Schweiger
et al., 1986). A low level of process design means that the leader does not design or
structure the process, but rather allows the group to determine both the mode of dialogue
among members and how to evaluate alternatives.
We suggest that TMT leaders can use process design to counteract the risk of failing to
generate or evaluate alternatives that arises in situations characterized by asymmetric
interests and information. Deliberate process design can help a TMT conduct a more
thorough analysis than would happen spontaneously when facing these asymmetries. For
example, creating subgroups forces the team to develop and consider different viewpoints,
and provides a forum where minority views are more accepted. By asking members to
explain and defend various alternatives in a structured debate, the leader enhances the
likelihood that private information will be shared. In addition, the leader can minimize
value-claiming behavior by dismantling natural coalitions in composing the subgroups, or
by forcing powerful advocates to consider and argue for options they may not have
endorsed initially. This may enhance the development of creative new options, as well as
the recognition and pursuit of value-creating opportunities.
This prescription is consistent with Janis’ (1982) observation that decision-making
practices employed during the Cuban Missile Crisis helped Kennedy’s team avoid
premature convergence. Following the Bay of Pigs fiasco, the Kennedy administration
developed techniques for employing subgroups and devil’s advocates. Reflecting on the
former decision, Kennedy and his advisers realized they had discouraged dissenting
opinions, marginalized those with minority views, and thereby converged prematurely on
a flawed plan of action. Janis thus advocated the use of subgroups to discuss and debate
alternatives, and he suggested assigning one or more team members to play devil’s
advocate.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 315
These structured decision-making procedures have been evaluated in a series of experi-
mental studies that suggest that formal, structured decision-making procedures are superior to
less structured techniques in certain situations (Priem et al., 1995; Schweiger et al., 1986).
Schweiger et al. (1986) demonstrated that both Dialectical Inquiry and Devil’s Advocacy
encouraged higher levels of critical evaluation, generated more alternatives, and led to higher
quality decisions than the less structured Consensus Method. Priem et al. (1995) demonstrated
that these structured methods promoted higher levels of team member satisfaction and
commitment, because individuals felt that they had a fair and legitimate opportunity to express
their views and disagree openly with one another. This research does not suggest that structured
techniques are uniformly superior to a consensus approach; conflicting findings have led
scholars to suggest that the effectiveness of each type of process depends on the nature of the
task (Murrell, Stewart, & Engel, 1993; Priem et al., 1995; Priem & Price, 1991). Consistent
with this, we propose that process design moderates the interaction effect of interest and
information asymmetries on TMT effectiveness.
Proposition 10: Process design reduces the negative effect on TMT effectiveness of the
interaction between situation-specific information and interest asymmetries.
Table 1 provides an overview of the process choices for each of the salient conditions we
identified above.
In summary, by examining the relationship between the stable team attributes studied in
TMT demographic research and the dynamic situation characteristics studied in supervisory
Fig. 2. Moderators of the relationship between situation-specific asymmetries and team effectiveness.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325316
leadership research, we suggest a new model of TMT effectiveness. The full model, shown
in Fig. 2, has the potential to explain conflicting findings from research that includes only
team-level variables. Fig. 2 summarizes our theoretical arguments by showing how both
team-level (stable) and situation-level (dynamic) variables moderate the relationship
between situation-specific (dynamic) asymmetries and team effectiveness. Thus, TMT
effectiveness is likely to vary across situations unless team process is carefully managed
by the leader.
6. Implications of the dynamic model of TMT effectiveness
Two essential leadership functions—behavioral style and interaction with subordinates
(supervisory leadership) and top executives’ influence on strategic choice and firm
performance (strategic leadership)—have been investigated in separate streams of research
(Boal & Hooijberg, 2000). The dynamic model presented in this paper suggests a need to
study these leadership functions together, and thus takes seriously the complementarity of
hierarchical and collaborative forms of leadership decision-making (Barnes & Kriger,
1986).
Our analysis thus contributes to building a cumulative field in leadership research (Hunt &
Dodge, 2001) by linking the leader’s process choices with TMT effectiveness—a key aspect
of strategic leadership effectiveness, which in turn is a key driver of organizational
effectiveness (Boal & Hooijberg, 2000). Fig. 3 depicts these relationships to place our model
in context. Because TMTs wrestle with strategic issues, effective teamwork can lead to better
strategic decisions and firm performance (Hambrick, 1994; Nadler, 1998). Consistent with
much leadership theory, our model (shown in Fig. 2) presents cross-level propositions
(Waldman & Yammarino, 1999), relating individual leaders’ process choices to group
effectiveness and organizational outcomes.
6.1. Implications for strategic leadership research
First, the construct of unstructured task streams and the resulting situation-specific
asymmetries they create suggest that group-level variables can only provide a limited
explanation of variation in TMT effectiveness. Static team characteristics, such as demo-
graphic composition, cannot lead uniformly to better or worse performance across all
situations. Our analysis suggests that the same team may perform more or less effectively in
Fig. 3. TMT effectiveness as an input to strategic leadership.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 317
different situations, such that findings related to demographic effects are likely to be
confounded by variability in distributions of information and interests within the team.
Second, if team effectiveness depends upon process choices that leaders make in each
situation, future research should include situational factors to explain TMT performance. This
work may help explain contradictory findings in the TMT literature.
Third, recognition of the effect of situational factors on team outcomes does not imply
that stable team characteristics are not important. Notably, we argue that psychological
safety and power centralization in TMTs moderate the relationship between situational
asymmetries and team effectiveness. Further theory development may consider effects of
stable TMT attributes on the likelihood that information or interest asymmetries will arise.
For example, team-level variables such as incentive structures or demographic heterogen-
eity are likely to influence the chances of situation-specific asymmetries occurring,
although they will not eliminate them completely. A compensation system that rewards
team members for the performance of individual business units rather than overall corporate
performance increases the chances of interest asymmetries. And, team longevity or tenure
may affect the chances of information asymmetry, because over time, teams can build
transactive memory (Moreland, 1999) in which they are more aware of what each member
knows.
6.2. Implications for supervisory leadership research
First, the propositions in this paper contribute to a long stream of research on
participative leadership (Avolio et al., 1996; Hunt, 1991; Schweiger & Leana, 1986; Yukl,
1989). Many authors have noted the limitations of advocating executive teamwork (e.g.,
Hambrick, 1994, 1995; Katzenbach, 1998). This paper suggests a new explanation for TMT
effectiveness and describes conditions under which process losses in TMTs are more likely,
as well as strategies for minimizing them, as summarized above in Table 1. In this way, we
have articulated process choices that can help achieve the ideal of shared (or distributed)
leadership (Gronn, 2002; Yukl, 1999). In contrast to the notion of a single heroic leader, the
construct of TMTs recognizes the potential value of shared leadership. Better process
choices constitute a step toward implementing that ideal.
Second, we specify process and outcome choices to help mitigate process losses in TMTs.
In so doing, we have disaggregated the notion of directive leadership (Tannebaum &
Schmidt, 1958; Vroom & Yetton, 1973). Normative decision theory presents a single
spectrum of directive versus participative leadership; we argue instead that ‘‘directive’’
leadership behavior may be found along three dimensions—outcome control, process
control, and process design. A highly directive approach involves high control on all three
dimensions; a nondirective approach, in contrast, engages a team in an unstructured process
without active leader intervention in the discussion, where closure is reached through
consensus.
We propose that these choices are independent. That is, a leader may choose to exert more
control with regard to structuring the debate but less control with regard to reaching closure,
and so on. However, we recognize that successful diagnosis requires a level of social
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325318
intelligence that not all leaders will have (Boal & Hooijberg, 2000), and successful execution
of the process choices similarly requires some behavioral complexity (Boal & Hooijberg,
2000; Hart & Quinn, 1993; Hunt, 1991). Our normative propositions thus require revisiting a
longstanding debate about whether leaders can exhibit genuine flexibility in adjusting their
styles contingent on the nature of the task (Fiedler, 1967; Hunt, 1991; Quinn, 1988; Yukl,
1989).
Past research provides preliminary evidence that leaders are indeed able to be
behaviorally flexible. Vroom and Yetton (1973) asked leaders to consider hypothetical
decision scenarios and to indicate how they would approach the situation; they found that
leaders believed they could exhibit flexibility without confirming that they actually do vary
their style across real situations. Frederickson and Mitchell (1984) examined hypothetical
decision scenarios with chief executives and found evidence that organizations employ
different types of decision-making process in different situations. Polley, Hare, and Stone
(1988) observed that senior members of small groups change behavior based on the group
process or issue. Katzenbach (1998) argued that leaders rely on teamwork at the top in
some situations but not in others, and Hart and Quinn (1993) provided empirical evidence
that CEOs employing a broad repertoire of behaviors enhanced organizational performance.
Finally, recent work in product development (Lewis et al., 2002) showed clearly that high
performing project leaders display a ‘‘flexible and complicated repertoire of activi-
ties. . .[and] go back and forth between styles as changes in project uncertainty occur’’
(p. 562).
Moving forward, a promising avenue for future empirical research is to examine whether
TMT leaders can change their degree of process and outcome control as the situation changes.
6.3. Model boundaries and future directions
This article examined leadership choices for shaping team process after a ‘‘task’’ has
been selected for attention. These choices necessarily take place within a larger context.
Before situation-specific process choices can be made, leaders (or, in some cases, members
of senior teams) must make choices about what tasks to address and who should be
involved in addressing them. This means that even team composition may vary across
different situations. In short, the leader is in a position to make structural choices in
addition to the process choices we have described. For example, the leader can decide
when and with whom ‘‘to team’’ at the top. Although it is beyond the scope of this paper
to suggest conditions for which a leader should use a team, future theoretical work should
develop a broader model, into which the model developed in this paper fits, to specify
these variables.
Unlike most work groups lower in the hierarchy, senior teams endogenously define the
specific task they will pursue next and, often, the membership of the group that will pursue a
given task. Membership choices affect when information and interest asymmetries arise and
thus indirectly contribute to process losses. The possibility of dynamic team composition has
been largely unexplored in the TMT literature, although recent empirical research indicates
that different teams may form at the top to address different types of strategic decisions
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 319
(Roberto, in press). Further theory development and empirical research are needed to develop
our understanding of how teams select and prioritize tasks, as well as how they vary the
composition of the decision-making group based upon the nature of the task at hand. This
paper is thus a starting point in building a richer concept of TMT effectiveness.
Finally, we do not intend to imply that the leader of a TMT necessarily must make one
set of process choices and remain wedded to those throughout a group decision process. It
is possible that leaders can continually update their assessments of the state of the team
and, over time, shift their approach to managing the team process. Suppose, for example,
the leader’s initial diagnosis for a situation is that there are significant information
asymmetries. As a result, the leader decides to intervene actively in the discussion to
surface and legitimize privately held information. Once the leader believes that the group
has surfaced all relevant information, it may make sense to shift approaches and intervene
less frequently in the discussion. Clearly, some shifts are easy to make, and others are more
difficult: for example, stepping back from high outcome control may be easier than
reversing the decision to exert low outcome control. Thus, research may wish to explore
the effect of real-time changes in the level of directive leadership in process design, process
control, and outcome control.
Our model ascribes the process leadership function to the CEO because the special
positional power and accountability of CEOs both motivates and gives them permission to
exercise process choices in TMTs. Future research, however, may explore the possibility
that multiple members of the TMT can share this leadership role, such that the process
design and facilitation responsibilities we have described in this paper can be distributed
within a team. This work could shed light on the specific circumstances in which shared
leadership of team process is more likely and more effective than process leadership by an
individual team leader.
This article provides a starting point for further theory development and empirical
research on TMT effectiveness emphasizing situational contingencies. We do not purport to
offer a complete explanation of variations in team performance. Instead, we offer new
insights and potential explanations for why senior teams often fail to fulfill expectations
and why they often experience certain modes of failure. Empirical research is needed to test
both the descriptive (related to failure modes) and normative (related to process choices)
propositions in this paper. In sum, the model described above invites investigation of the
‘‘black box’’ of senior teams, to examine the dynamic internal processes that are generally
ignored in the study of effects of TMT attributes on organizational outcomes.
Acknowledgements
The authors contributed equally to this article. We are grateful to Max Bazerman, Nancy
Beaulieu, Ralph Biggadike, Joe Bower, David Garvin, Rosabeth Moss Kanter, Hannah
Riley, and Mike Tushman for feedback on an earlier version of this article, and to members
of the Senior Teams Workshop and the Negotiations, Organizations, and Markets Seminar at
the Harvard Business School for ideas and inspiration. The final article benefited greatly
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325320
from helpful comments provided by three anonymous reviewers and Jerry Hunt at The
Leadership Quarterly.
References
Allison, G. T. (1971). The essence of decision: explaining the Cuban missile crisis. Boston, MA: Little Brown.
Amason, A. C. (1996). Distinguishing the effects of functional and dysfunctional conflict on strategic decision
making. Academy of Management Journal, 39(1), 123–148.
Ancona, D., & Nadler, D. (1989). Top hats and executive tales. Sloan Management Review, 31(1), 19–28.
Argyris, C. (1993). Knowledge for action: a guide to overcoming barriers to organizational change. San Fran-
cisco: Jossey-Bass.
Asch, S. E. (1951). Effects of group pressure upon the modification and distortion of judgments. In H. Guetzgow
(Ed.), Groups, leadership and men (pp. 177–190). Pittsburgh, PA: Carnegie.
Avolio, B. J., Jung, D. I., Murry, W., & Sivasubramaniam, N. (1996). Building highly developed teams: focusing
on shared leadership processes, efficacy, trust, and performance. In M. Beyerlein, D. Johnson, & S. Beyerlein
(Eds.), Advances in interdisciplinary studies of work teams, vol. 3 (pp. 173–209). Stamford, CT: JAI Press.
Bales, R. (1954). In conference. Harvard Business Review, 32(2), 44–50.
Bales, R. (1988). A new overview of the SYMLOG system: measuring and changing behavior in groups. In
R. Polley, A. Hare, & P. Stone (Eds.), The SYMLOG practitioner: applications of small group research
(pp. 319–344). New York: Praeger.
Bantel, K., & Jackson, S. (1989). Top management and innovations in banking: does the composition of the top
team make a difference? Strategic Management Journal, 10, 107–124.
Barnes, L. B., & Kriger, M. P. (1986). The hidden side of organizational leadership. Sloan Management
Review, 28(1), 15–26.
Baron, R., Vandello, J., & Brunsman, B. (1996). The forgotten variable in conformity research: impact of task
importance on social influence. Journal of Personality and Social Psychology, 71, 915–927.
Bauman, R., Jackson, P., & Lawrence, J. (1997). From promise to performance, a journey of transformation at
SmithKline Beecham. Boston, MA: Harvard Business School Press.
Bazerman, M. (1998). Judgment in managerial decision making (4th ed.). New York: Wiley.
Bazerman, M., Mannix, E., & Thompson, L. (1988). Groups as mixed-motive negotiations. In E. J. Lawlor, & B.
Markovsky (Eds.), Advances in group processes: a research annal, vol. 5 (pp. 195–216). Greenwich, CT: JAI
Press.
Bazerman, M., & Neale, M. A. (1992). Negotiating rationally. New York: Free Press.
Boal, K. B., & Hooijberg, R. J. (2000). Strategic leadership research: moving on. Yearly Review of Leadership: A
Special Issue of The Leadership Quarterly, 11, 515–550.
Bower, J. L. (1970). Managing the resource allocation process. Boston, MA: Harvard Business School Press.
Bower, J. L. (1998). Process research on strategic decisions. In V. Papadakis, & P. Barwise (Eds.), Strategic
decisions (pp. 17–33). Boston, MA: Kluwer Academic Publishing.
Cannon-Bowers, J. A., Salas, E., & Converse, S. (1993). Shared mental models in expert team decision making. In
N. J. Castellan (Ed.), Individual and group decision making (pp. 241–246). Hillsdale, NJ: Erlbaum.
Child, J. (1972). Organizational structure, environment, and performance: the role of strategic choice. Sociol-
ogy, 6, 1–22.
Clark, K., & Wheelwright, S. C. (1992). Organizing and leading ‘‘heavyweight’’ development teams. California
Management Review, 34(3), 201–215.
Cyert, R. M., & March, J. G. (1963). The behavioral theory of the firm. Cambridge, MA: Blackwell.
Dougherty, D. (1992). Interpretive barriers to successful product innovation in large firms. Organization Sci-
ence, 3(2), 179–202.
Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quar-
terly, 44(4), 350–383.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 321
Edmondson, A. (2002). The local and variegated nature of learning in organizations. Organization Science, 13(2),
128–146.
Edmondson, A., Bohmer, R., & Pisano, G. (2000). Learning new technical and interpersonal routines in operating
room teams: the case of minimally invasive cardiac surgery. In B. Mannix, M. Neale, & T. Griffith (Eds.),
Research on groups and teams (pp. 29–51). Greenwich, CT: JAI Press.
Edmondson, A., Bohmer, R., & Pisano, G. (2001). Disrupted routines: team learning and new technology
implementation in hospitals. Administrative Science Quarterly, 46(4), 685–716.
Eisenhardt, K. M. (1989). Making fast strategic decisions in high-velocity environments. Academy of Management
Journal, 32(4), 543–576.
Eisenhardt, K. M., & Bourgeois, L. J. (1988). Politics of strategic decision making in high-velocity environments:
toward a midrange theory. Academy of Management Journal, 31(4), 737–770.
Eisenhardt, K. M., & Schoonhoven, C. (1990). Organizational growth: linking team founding, strategy, environ-
ment, and growth among U.S. semiconductor ventures, 1978–88. Administrative Science Quarterly, 35(3),
504–529.
Elangovan, A. (1995). Managerial third-party dispute resolution: a prescriptive model of strategy selection.
Academy of Management Review, 20(4), 800–830.
Elangovan, A. (1998). Managerial intervention in organizational disputes: testing a prescriptive model of strategy
selection. International Journal of Conflict Management, 9(4), 301–335.
Evans, M. G. (1970). The effects of supervisory behavior on the path–goal relationship. Organizational Behavior
and Human Performance, 5, 277–298.
Fiedler, F. E. (1967). A theory of leadership effectiveness. New York: McGraw-Hill.
Field, R. (1982). A test of the Vroom–Yetton normative model of leadership. Journal of Applied Psychol-
ogy, 67(5), 523–532.
Finkelstein, S. (1988). Managerial orientations and organizational outcomes: the moderating roles of managerial
discretion and power. Unpublished doctoral dissertation, Columbia University, New York.
Finkelstein, S. (1992). Power in top management teams: dimensions, measurement, and validation. Academy of
Management Journal, 35(3), 505–538.
Finkelstein, S., & Hambrick, D. C. (1990). Top management team tenure and organizational outcomes: the
moderating role of managerial discretion. Administrative Science Quarterly, 35(3), 484–503.
Finkelstein, S., & Hambrick, D. C. (1994). Strategic leadership: top executives and their effects on organizations.
Minneapolis, MN: West Publishing.
Fisher, R., & Ury, R. (1991). Getting to yes. New York: Penguin.
Folger, R., & Konovsky, M. (1989). Effects of procedural and distributive justice. In R. Lewicki, B. Sheppard, &
M. Bazerman (Eds.), Research on negotiation in organizations (pp. 57–80). Greenwich, CT: JAI Press.
Frederickson, J. W., & Mitchell, T. R. (1984). Strategic decision processes: comprehensiveness and performance
in an industry with an unstable environment. Academy of Management Journal, 27(3), 399–423.
Garvin, D., & Roberto, M. (2001). What you don’t know about making decisions. Harvard Business Review,
79(8), 108–116.
Gronn, P. (2002). Distributed leadership as a unit of analysis. Leadership Quarterly, 13, 423–451.
Hackman, J. R. (1987). The design of work teams. In J. Lorsch (Ed.), Handbook of organizational behavior
(pp. 315–342). Englewood Cliffs, NJ: Prentice-Hall.
Hackman, J. R. (1990). Groups that work (and those that don’t). San Francisco: Jossey-Bass.
Hackman, J. R. (1998). Why groups don’t work. In R. S. Tindale, & L. Heath (Eds.), Theory and research
on small groups: social psychological applications to social issues, vol. 4 (pp. 245–267). New York:
Plenum.
Hambrick, D. C. (1994). Top management groups: a conceptual integration and reconsideration of the team label.
In B. M. Staw, & L. L. Cummings (Eds.), Research in organizational behavior (pp. 171–214). Greenwich,
CT: JAI Press.
Hambrick, D. C. (1995). Fragmentation and other problems CEOs have with their top management teams.
California Management Review, 37(3), 110–127.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325322
Hambrick, D. C. (1998). Corporate coherence and the top management team. In D. Hambrick, D. Nadler, & M.
Tushman (Eds.), Navigating change (pp. 123–140). Boston, MA: Harvard Business School Press.
Hambrick, D. C., Cho, T. S., & Chen, M. J. (1996). The influence of top management team heterogeneity on
firms’ competitive moves. Administrative Science Quarterly, 41(4), 659–684.
Hambrick, D., & Mason, P. (1984). Upper echelons: the organization as a reflection of its top managers. Academy
of Management Review, 9, 193–206.
Harrison, E. F. (1996). The managerial decision-making process (4th ed.). Boston, MA: Houghton-Mifflin.
Hart, S. L., & Quinn, R. E. (1993). Roles executives play: CEOs, behavioral complexity and firm performance.
Human Relations, 46(5), 543–574.
Hersey, P., & Blanchard, K. (1969). Life cycle theory of leadership. Training and Development Journal, 23(2),
26–34.
Hickson, D. J., Wilson, D. C., Cray, D., Mallory, G. R., & Butler, R. J. (1986). Top decision: strategic decision
making in organizations. San Francisco: Jossey-Bass.
Hollenbeck, J. R., Ilgen, D. R., LePine, J. A., Colquitt, J. A., & Hedlund, J. (1998). Extending the multilevel
theory of team decision making: effects of feedback and experience in hierarchical teams. Academy of Manage-
ment Journal, 41(3), 269–282.
Hollenbeck, J. R., Ilgen, D. R., Sego, D., Hedlund, J., Major, D. A., & Phillips, J. (1995). The multilevel theory of
team decision making: decision performance in teams incorporating distributed expertise. Journal of Applied
Psychology, 80(2), 292–316.
Hovland, C., & Weiss, W. (1951). The influence of source credibility on communication effectiveness. Public
Opinion Quarterly, 15, 635–650.
Hunt, J. G. (1991). Leadership: a new synthesis. Newbury Park, CA: Sage Publications.
Hunt, J. G., & Dodge, G. E. (2001). Leadership deja vu all over again. Leadership Quarterly, 11(4), 435–458.
Janis, I. L. (1982). Victims of groupthink (2nd ed.). Boston, MA: Houghton-Mifflin.
Janis, I. L., & Mann, L. (1977). Decision making: a psychological analysis of conflict, choice, and commitment.
New York: Free Press.
Janofsky, M. (1993, March 27). At RJR, two heads will act as one. The New York Times, pp. 35, 46.
Jehn, K. A. (1995). A multimethod examination of the benefits and detriments of intragroup conflict. Admin-
istrative Science Quarterly, 40(2), 256–282.
Kanter, R. M. (1979). Power failure in management circuits. Harvard Business Review, 57(4), 65–72.
Karambayya, R., & Brett, J. M. (1989). Managers handling disputes: third-party roles and perceptions of fairness.
Academy of Management Journal, 32(4), 687–704.
Katzenbach, J. R. (1998). Teams at the top. Boston, MA: Harvard Business School Press.
Keck, S., & Tushman, M. (1993). Environmental and organizational context and executive team structure. Acad-
emy of Management Journal, 36(6), 1314–1344.
Kim, W. C., & Mauborgne, R. (1997). Fair process: managing in the knowledge economy. Harvard Business
Review, 75(4), 65–75.
Knight, D., Pearce, C. L., Smith, K. G., Olian, J. D., Sims, H. P., Smith, K. A., & Flood, P. (1999). Top
management team diversity, group process, and strategic consensus. Strategic Management Journal, 20(5),
445–465.
Korsgaard, M., Schweiger, D., & Sapienza, H. (1995). Building commitment, attachment, and trust in strategic
decision-making teams: the role of procedural justice. Academy of Management Journal, 38(1), 60–84.
Larson, J., Christensen, C., Abbott, A., & Franz, T. (1996). Diagnosing groups: charting the flow of information in
medical decision making teams. Journal of Personality and Social Psychology, 71, 315–330.
Larson, J., Foster-Fishman, P., & Franz, T. (1998). Leadership style and the discussion of shared and unshared
information in decision-making groups. Personality and Social Psychology Bulletin, 24(5), 482–495.
Larson, J. R., Foster-Fishman, P. G., & Keys, C. B. (1994). Discussion of shared and unshared information in
decision-making groups. Journal of Personality and Social Psychology, 67(3), 446–461.
Lax, D., & Sebenius, J. (1986). The manager as negotiator: bargaining for cooperation and competitive gain.
New York: Free Press.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 323
Levine, J. M. (1989). Reaction to opinion deviance in small groups. In P. B. Paulus (Ed.), Psychology of group
influence (pp. 187–232). Hillsdale, NJ: Erlbaum.
Lewis, M. W., Welsh, M. A., Dehler, G. E., & Green, S. G. (2002). Product development tensions: exploring
contrasting styles of product management. Academy of Management Journal, 45(3), 546–564.
Maier, N. (1961). Assets and liabilities in group problem solving: the need for an integrative function. Psycho-
logical Review, 74, 239–249.
Mannix, E. (1989). Coalitions in the organizational context: a social dilemmas perspective. Unpublished dis-
sertation. University of Chicago.
Mannix, E. (1993). Organizations as resource dilemmas: the effects of power balance on coalition formation in
small groups. Organizational Behavior and Human Decision Processes, 55(1), 1–22.
Mannix, E., & Neale, M. (1993). Power imbalance and the pattern of exchange in dyadic negotiation. Group
Decision and Negotiation, 2(2), 119–133.
McAlister, L., Bazerman, M., & Fader, P. (1986). Power and goal setting in channel negotiations. Journal of
Marketing Research, 23(3), 238–263.
McClintock, C., Messick, D., Kuhlman, D., & Campos, F. (1973). Motivational bases of choice in three-choice
decomposed games. Journal of Experimental Social Psychology, 9, 572–590.
Miller, C. C., Burke, L. M., & Glick, W. H. (1998). Cognitive diversity among upper-echelon executives:
implications for strategic decision processes. Strategic Management Journal, 19(1), 39–58.
Mintzberg, H. (1983). Power in and around organizations. Englewood Cliffs, NJ: Prentice-Hall.
Moreland, R. L. (1999). Transactive memory: learning who knows what in work groups and organizations. In
L. L. Thompson, J. M. Levine, & D. M. Messick (Eds.), Shared cognition in organizations: the management
of knowledge (pp. 3–31). Mahwah, NJ: Erlbaum.
Murmann, J. P., & Tushman, M. (1997). The effects of executive team characteristics and organizational context
on organizational responsiveness to environmental shock. Working Paper, Columbia Business School.
Murray, A. (1989). Top management group heterogeneity and firm performance. Strategic Management Jour-
nal, 10, 125–141.
Murray, E. (1978). Strategic choice as a negotiated outcome. Management Science, 24(9), 960–972.
Murrell, A., Stewart, A., & Engel, B. (1993). Consensus versus devil’s advocacy: the influence of decision process
and task structure on strategic decision making. Journal of Business Communication, 30(4), 399–414.
Nadler, D. A. (1996). Managing the team at the top. Strategy and Business, 2, 42–51.
Nadler, D. A. (1998). Leading executive teams. In D. Nadler, J. Spencer, & Associates (Eds.), Executive teams
(pp. 3–20). San Francisco: Jossey-Bass.
Nemeth, C. (1997). Managing innovation: when less is more. California Management Review, 40(1), 59–74.
Nemeth, C., & Wachtler, J. (1974). Creating perceptions of consistency and conformity: a necessary condition for
minority influence. Sociometry, 37, 529–540.
Papadakis, V. M., Lioukas, S., & Chambers, D. (1998). Strategic decision-making processes: the role of manage-
ment and context. Strategic Management Journal, 19(2), 115–147.
Pfeffer, J. (1981). Power in organizations. Marshfield, MA: Pitman.
Pfeffer, J. (1992). Managing with power. Boston, MA: Harvard Business School Press.
Pettigrew, A. M. (1973). The politics of organizational decision making. London: Tavistock.
Pettigrew, A. M. (1992). On studying managerial elites. Strategic Management Journal, 13, 163–182.
Polley, R., Hare, A., & Stone, P. (Eds.) (1988). The SYMLOG practitioner: applications of small group research.
New York: Praeger.
Priem, R. L., Harrison, D. A., & Muir, N. K. (1995). Structured conflict and consensus outcomes in group
decision making. Journal of Management, 21(4), 691–710.
Priem, R. L., & Price, K. H. (1991). Process and outcome expectations for the dialectical inquiry, devil’s advocacy
and consensus techniques of strategic decision making. Group and Organization Studies, 16(2), 206–225.
Quinn, R. E. (1988). Beyond rational management: mastering the paradoxes and competing demands of high
performance. San Francisco: Jossey-Bass.
Raiffa, H. (1982). The art and science of negotiation. Cambridge, MA: Harvard University Press.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325324
Roberto, M. (2000). Strategic decision-making processes: achieving efficiency and consensus simultaneously.
Unpublished doctoral dissertation. Harvard Business School.
Roberto, M. (in press). The stable core and dynamic periphery in top management teams. Management
Decision.
Ross, J., & Staw, B. (1986). Expo 86: an escalation prototype. Administrative Science Quarterly, 31(2), 274–297.
Ross, J., & Staw, B. (1993). Organizational escalation and exit: lessons from the Shoreham nuclear power plant.
Academy of Management Journal, 36(4), 701–732.
Russo, J. E., & Schoemaker, P. (2002). Winning decisions. New York: Currency.
Schachter, S. (1951). Deviation, rejection, and communication. Journal of Abnormal and Social Psychology, 46,
190–207.
Schwartz, R. M. (1994). The skilled facilitator: practical wisdom for developing effective groups. San Francisco:
Jossey-Bass.
Schweiger, D. M., & Leana, C. (1986). Participation in decision-making. In E. Locke (Ed.), Generalizing from
laboratory to field settings (pp. 147–166). Lexington, MA: Lexington Books.
Schweiger, D. M., Sandberg, W. R., & Ragan, J. W. (1986). Group approaches for improving strategic
decision making. Academy of Management Journal, 29(1), 51–71.
Seers, A. (1996). Better leadership through chemistry: toward a model of emergent shared team leadership. In
M. M. Beyerlein, D. A. Johnson, & S. T. Beyerlein (Eds.), Advances in interdisciplinary studies of work
teams, vol. 3 (pp. 145–172). Stamford, CT: JAI Press.
Shapiro, D. (1993). Reconciling theoretical differences among procedural justice researchers by re-evaluating
what it means to have one’s views ‘‘considered’’: implications for third-party managers. In R. Cropanzano
(Ed.), Justice in the workplace: approaching fairness in human resource management (pp. 231–262). Hill-
sdale, NJ: Erlbaum.
Smith, K. G., Smith, K. A., Olian, J., Sims, H., O’Bannon, D., & Scully, J. (1994). Top management
demography and process: the role of social integration and communication. Administrative Science Quar-
terly, 39(3), 412–438.
Spreitzer, G. M. (1995). Psychological empowerment in the workplace: dimensions, measurement, and validation.
Academy of Management Journal, 38(5), 1442–1465.
Stasser, G. (1999). The uncertain role of unshared information in collective choice. In L. Thompson, J. Levine, &
D. Messick (Eds.), Shared cognition in organizations (pp. 49–69). Mahwah, NJ: Erlbaum.
Stasser, G., & Davis, J. J. (1981). Group decision making and social influence: a social interaction model.
Psychological Review, 88, 523–551.
Stasser, G., & Titus, W. (1985). Pooling of unshared information in group decision making: biased information
sampling during discussion. Journal of Personality and Social Psychology, 48, 1467–1478.
Steiner, I. D. (1972). Group process and productivity. New York: Academic Press.
Tannebaum, R., & Schmidt, W. H. (1958). How to choose a leadership pattern. Harvard Business Review, 36(2),
95–101.
Taras, D. G. (1991). Breaking the silence: differentiating crises of agreement. Public Administration Quar-
terly, 14(4), 401–419.
Thibault, J., & Walker, J. (1975). Procedural justice: a psychological analysis. Hillsdale, NJ: Erlbaum.
Tushman, M. (1977). A political approach to organizations: a review and rationale. Academy of Management
Review, 2(2), 206–216.
Vroom, V., & Jago, A. (1988). The new leadership: managing participation in organizations. Englewood Cliffs,
NJ: Prentice-Hall.
Vroom, V. H., & Yetton, P. W. (1973). Leadership and decision making. Pittsburgh, PA: University of Pitts-
burgh Press.
Wagner, W., Pfeffer, J., & O’Reilly, C. (1984). Organizational demography and turnover in top management
teams. Administrative Science Quarterly, 29(1), 74–92.
Waldman, D. A., & Yammarino, F. J. (1999). CEO charismatic leadership: levels of management and levels of
analysis. Academy of Management Review, 24(2), 266–285.
A.C. Edmondson et al. / The Leadership Quarterly 14 (2003) 297–325 325
Walton, R., & McKersie, R. (1965). A behavioral theory of labor negotiations. New York: McGraw-Hill.
Watkins, M. (2000). Negotiation analysis: a synthesis. (Harvard Business School Note #800-316). Boston:
Harvard Business School Publishing.
Webne-Behrman, H. (1998). Practice of facilitation: managing group process and solving problems. London:
Quorum Books.
Williams, K., & O’Reilly, C. (1998). Demography and diversity in organizations: a review of 40 years of
research. In B. Staw, & R. Sutton (Eds.), Research in organizational behavior (pp. 77–140). Greenwich,
CT: JAI Press.
Wooldridge, B., & Floyd, S. W. (1990). The strategy process, middle management involvement, and organiza-
tional performance. Strategic Management Journal, 11(3), 231–241.
Yukl, G. (1989). Managerial leadership: a review of theory and research. Journal of Management, 15(2),
251–289.
Yukl, G. (1999). An evaluation of conceptual weaknesses in transformational and charismatic leadership theories.
Leadership Quarterly, 10(2), 285–305.
Zimbardo, P., & Leippe, M. (1991). Psychology of attitude change and social influence (3rd ed.). Philadelphia:
Temple Press.